The optimal detection of ionizing radiation in two dimensions is the central problem in computed tomography, digital radiography, nuclear medicine imaging and related disciplines. Many different types of detectors (e.g. non-electronic, analog electronic and digital electronic detectors) have been used with varying degrees of success in these fields. In general, many compromises have been made among the various imaging and non-imaging parameters of prior art detectors in developing operational systems.
It has been recognized for some time that there is no fundamental impediment to the replacement of film and other non-digital radiographic techniques with digital detection methods, and that the basic problems was one of developing a suitable detector and data acquisition system (DAS). See generally Foley & DiBianca, "Computed Radiography" in Radiology of the Skull and Brain: Technical Aspects of Computed Tomography at chapter 128, pages 4312-25 (Newton & Potts 1981). One proposed digital radiography system 10 is shown schematically in FIG. 1. A source of x-rays 12 radiates x-radiation toward a collimator 14. An approximately square aperture 16 is defined in collimator 14 to direct an approximately square x-radiation distribution (i.e. wide area beam) 18 toward a patient 20. X-rays produced by source 12 which do not pass through aperture 16 are blocked by collimator 14 (which preferably is made of a very dense material such as lead or the like) and therefore do not strike patient 20. The portions of wide area beam 18 passing through patient 20 travel further to strike an approximately square detector 22 positioned behind patient 20. The intensity of the radiation exiting patient 20 along any path depends on the integrated x-ray attenuation coefficient of the patient along that path.
Detector 22 has a side 23 having a length L of approximately 50 centimeters to match the size of beam 18 after it is passed through patient 20 (because x-ray source 12 resembles a point source, wide-area beam 18 spreads as it travels away from collimator 14). Detector 22 produces signals corresponding to the intensity of the x-radiation at the various points in the two dimensions of the detector which can be further processed by conventional techniques to obtain an image of the projection of the density of patient 20 onto the two-dimensional plane of the detector.
Detector 22 comprises a plurality of discrete detecting elements 24 arranged in a two-dimensional coordinate array. If the desired limiting spatial resolution of system 10 is five line pairs/mm, each detecting element 24 would have a square dimension s of length 0.125 mm (see FIG. 1A) for a magnification of 1.25. Wide-area detector 22 would then contain n.sup.2 elements with n equal to 4000 (for a total of 16 million discrete elements).
It is not feasible at present to construct such a large detector 22 with so many discrete electronic elements 24. Therefore, devices with continuous detectors have been proposed and evaluated for digital radiography. Examples of such continuous detectors are large area image intensifiers, see Rowlands et al, III Optical Instrumentation In Medicine (SPIE, Washington 1984), and photostimulable phosphor screens, see Sonada et al, 148 Radiology 833 (1983). Such continuous detectors, however, have problems with scattered radiation acceptance, limited detective quantum efficiency, light spreading and other difficulties which limit system performance.
The above-mentioned problems of continuous detectors may be largely overcome along with the problems of the mechanical and electronic complexity of an n.sup.2 discrete element detector by utilizing a thin scanning fan beam of radiation and an n-by-one element detector. See, for example, U.S. Pat. No. 3,983,398 to Boyd (1976); U.S. Pat. No. 4,075,492 to Boyd et al (1978); DiBianca et al, 133 Radiology 231 (1979); and DiBianca et al, 15 Inv. Radiology 220 (1980). An example of a known scanning fan beam radiography system 30 is shown schematically in FIG. 2. A collimator 14 defines a slot 32 through which x-radiation produced by x-radiation source 12 is directed. The resulting fan-shaped beam 34 is directed through the patient 20 onto an n-by-one element detector 36 comprising n discrete detecting elements 24 arranged in a linear array along an x-coordinate axis. The fan-shaped beam 34 is scanned over the portions of patient 20 of interest by moving collimator 14. Detector 36 is moved simultaneously in a direction perpendicular to the plane of beam 34 (such as by linearly translating an arm, not shown, on which collimator 14 and the detector are commonly mounted) so that beam 34 is always incident on detector 36. A focused grid collimator may be interposed between patient 20 and detector 36 for collimating the radiation penetrating the patient onto the detector. The position in an x coordinate direction of an element 24 of detector 36 producing a signal indicates the position in the x direction of the z-radiation causing the signal to be produced by the element. The position of detector 36 in a z-coordinate direction (i.e., scanning direction) perpendicular to the x direction at the time the signal is produced indicates the position in the z direction of the x-radiation producing that signal.
Unfortunately, a number of difficulties are also involved with thin scanning fan beam system 30. The very thin (approximately 0.1 mm) x-radiation fan beam 34 required for a resolution of five line pairs/mm uses the x-ray flux produced by source 12 very inefficiently and thus produces either excessive image noise or unacceptably long scan times and excessive x-ray source (tube) loading. In addition, the focal spot penumbra of system 30 seriously degrades spatial resolution of the system in the scanning (z) direction.
One compromise solution is to use an n-by-m detector geometry with, for example, m=.sqroot.n together with a thick fan beam, sometimes called a "strip" beam. Such a system is disclosed in Wang et al, XII Optical Instrumentation In Medicine 250 (SPIE, Washington, 1984). Although such a device might overcome some of the physical problems discussed above, it is unclear how a discrete element detector and data acquisition system with the 4000 by 64 (i.e. 256,000) channels necessary to obtain useful resolution could be constructed in practice.
Xenon gas ionization detectors have been used successfully in a number of third generation commercial and experimental computed tomography and digital radiography systems. A typical xenon detector 50 for use in digital radiography is illustrated in FIG. 3. Detector 50 comprises a high voltage plate 52 and a collection plate 54 disposed parallel to the high-voltage plate. The space 56 between plates 52 and 54 is filled with a pressurized quantity of high atomic number ionizable gas such as xenon. Space 56 comprises a detection volume in which ionizing events are produced in the xenon gas by x-rays 59 incident thereto.
A strong electric field is produced between plates 52 and 54 b applying a high electric potential across the plates. Positive ions produced in space 56 by absorption of incident x-rays are attracted to collection plate 54, and electrons are attracted to high-voltage plate 52 Since the number of ion-electron pairs produced in space 56 is proportional to the intensity of the radiation incident on detector 50, the current flowing in collection plate 54 can be used as an indicator of incident x-ray intensity (or the transmissivity of an object interposed between the x-ray source and detector 50).
Plate 54 comprises a circuit board 57 etched to form an array of conductive collection electrodes 58. The collection electrodes 58 are focused on the source of x-rays (i.e. an x-ray tube focal spot) and therefore may be wider at the rear 60 of detector 50 than at the front 62 of the detector. A respective detection volume is defined by each of collection electrodes 58, the detection volume having a length L and width W defined by the and having a height H defined by the separation between collection plate 58 and high-voltage plate 52.
In the detector 50 shown in FIG. 3, there are no separating elements between individual detection volumes. This makes the construction of an array of elements with submillimeter widths W relatively straight-forward, permitting an n-by-one detector which has a large number of detection volumes per unit length to be constructed. The absence of separating elements between detection volumes may lead to degradation of spatial resolution due to cross-talk between adjacent detection volumes. However, at high gas pressures, the cross-talk for collection electrodes 58 having a width of 0.5 mm falls to less than 10% because the gas itself restricts charge carriers formed in one detection volume from moving to adjacent detection volumes. See Fenster et al, "Characteristics of A Linear Xenon Detector Array For Scanned Projection Radiography", Proceedings of the AAPM Summer School 214-44 (1984); Drost et al, "A Xenon Ionization Detector For Digital Radiography", Vol. 9, No. 2, Med. Phys. 224-30 (1982); and Rutt et al, "A Xenon Ionization Detector For Scanned Projection Radiography: Theoretical Considerations", Vol. 10, No. 3 Med. Phys. 284-92 (1983).
An analysis of the theory and performance of a xenon gas ionization detector wherein alternating planar high-voltage and collector electrodes define gaps in which charge carriers are produced by x-radiation entering through the front window of a hermetically sealed housing is disclosed in Peschmann, "Xenon Gas Ionization Detectors" in Radiology of the Skull and Brain: Technical Aspects of Computer Tomography, Section 3, pages 4112-26 (Newton & Potts 1981) and U.S. Pat. No. 4,031,396 to Whetten et al. In the system discussed by Peschmann, the x-radiation beam is pulsed and the resulting charges collected by the detection elements are integrated over time to decrease signal quality degradation produced by the natural fluctuations of the x-radiation beam intensity.
Parallel-plate gas ionization chambers have been used for medical imaging in other ways in the past. Johns et al, "Gas Ionization Methods OF Electrostatic Image Formation in Radiography", 47 British Journal of Radiology 519-29 (1974) discloses a wide-gap chamber containing a pressurized high-Z gas which is ionized by a pencil beam of x-radiation. Johns et al discuss the radial distribution of produced charge carriers and the effect of ion diffusion on ion detection.
U.S. Pat. No. 4,286,158 to Charpak et al (1981) discloses an ion chamber using photomultiplier tubes to detect the positions and brightnesses of scintillations produced by the formation of secondary photons to ascertain radiation spatial distribution and intensity. U.S. Pat. No. 4,317,038 to Charpak (1982) discloses a similar ion chamber operated as a multi-wire proportional chamber. In this latter device, flat grids disposed in the chamber induce charge multiplication from photo-electrons produced by x-radiation absorbed by a noble gas within the Chamber. The multiplied charges are detected by a set of electrode wires.
U.S. Pat. No. 4,320,299 to Bateman et al (1982) discloses an ionization chamber with a position-sensitive multi-wire array on which an electrical charge is induced by charge multiplication of electrons and positive ions. U.S. Pat. No. 4,485,307 to Osborne et al (1984) discloses a similar spatial detection gas ionization chamber including detector wires formed in a crossed mesh pattern.
U.S. Pat. No. 4,057,728 to Peschmann et al (1977) teaches a gas ionization chamber adapted for x-ray detection which includes an insulating foil imaging plane displaced in the longitudinal direction of the chamber by a variable amount dependent on the x-ray angle of incidence. A follower control system controlled by the x-ray angle of incidence moves a carriage on which the insulating foil is mounted.
U.S. Pat. No. 3,963,924 to Boag et al (1976) discloses a xenon gas ionization chamber including electrodes with spherically curved surfaces. The effect of the curved surfaces is to maintain the x-ray beam passing through the object to be imaged normal to the electrode surfaces. In this way, the lines of force of the collecting field are always parallel to the quantum paths of the ions formed by the incident x-rays.
Gas ionization chambers have been used for many years for a variety of applications other than medical imaging. For instance, gas ionization drift chambers are used in physics for determining the path of a particle in 3-dimensional space. When a high-energy nuclear particle travels through a gaseous medium within a chamber, it leaves a track of charge carriers (ions). A plane of wires disposed in the chamber produces an electric field to attract the charge carriers so produced As the charge carriers approach the plane of wires, the intensity of the electric field increases the velocity of the charge carries, causing charge multiplication (avalanching) and inducing current to flow in the wires. Electronics connected to the wires measures the current flowing in the wires with respect to time. The wires in the plane are formed into a grid to permit the x and y coordinates of the ionization events to be ascertained. The arrival time of the charge carriers at the plane of wires determines the position of the ion track in the z coordinate direction. See, for example, "The Time Projection Chamber", American Institute of Physics Conference Proceedings No. 108 (New York 1984); U.S. Pat. No 4,179,608 to Walenta (1979).
A serious drawback of conventional gas ionization radiography detectors is that the maximum resolution obtainable is limited to the distance between the electrodes establishing the electric field. As the electrode spacing is decreased, the detector uses radiation less efficiently (due to the higher ratio of electrode volume to detection volume) and detective quantum efficiency decreases. Moreover, minimum electrode spacing is limited by mechanical factors and in any event cannot be made less than the spacing necessary to ensure that no electrical arcing between electrodes occurs. Thus, high resolutions are presently difficult or impossible to obtain in practice with this type of detector.
Perhaps the major drawback of gas ionization radiography detectors, however, is their relatively slow recovery time. The time it takes for charge carriers formed on the side of the chamber opposite to the collection grid to drift through the chamber and reach the collection grid depends upon the size of the chamber, the ion mobility of the gas within the chamber, and the electric field intensity. Typically, it takes ionic charge carries a few milliseconds to traverse the chamber and reach the collection grid. New ionization events occurring during this time period (which is relatively long on an atomic scale) cannot be distinguished from an earlier event, and will cause erroneous results. For this reason, radiation sources are often operated in the pulse mode with times between pulses greater than the time required for charge carriers to completely traverse the chamber. See, for example, U.S. Pat. No. 4,301,368 to Riihimaki (1981) (proportional mode gas ionization chamber). Even in pulse mode operation, it is not possible to distinguish between plural ionization events occurring closely together in time in the same detection element.
Moreover, charge carriers located anywhere in an ionization chamber continuously induce a charge on the collection electrode of the chamber while they are drifting toward the electrode. Consider the formation of a single ion pair somewhere in an ionization chamber. Under the influence of the electric field, the positive ion and the electron (e.sup.-) separate, each drifting towards an oppositely-charged electrode. One might believe that when a charged particle arrives on the collector plate, the potential of the collector plate changes by -e/C (where C is the total capacitance of the collector plate). This view is not correct, however, because it neglects the induction effects which the two ions have been exerting on both plates since the time of creation of the ion pair.
At time t after the ion pair is formed, the positive and negative charge carriers induce charges -q.sub.+ (t) and -q.sub.- (t) on the positive electrode. The potential P(t) of the positive electrode, originally zero, becomes ##EQU1## (assuming the time constant of the electrode is long compared to t). The charge induced on the other electrode of an infinite two-electrode system is complementary. The current pulses flowing in the two electrodes are thus identical in shape and amplitude although different in sign.
At the instant of formation of the ion pair, the following relation must hold true: EQU q.sub.- (0)=-q.sub.+ (0), (2) EQU So that EQU P(0)=0 (3)
When the negative ion is collected at time t.sub.1, all of its charge must be induced on the collection electrode. Therefore, EQU q-(t.sub.1)=-e (4)
(assuming the negative ion is collected first). At the instant when the simple view would suggest that the potential of the positive electrode should be -e/C, it is thus actually ##EQU2##
At time t.sub.2 &gt;t.sub.1, when the positive ion is collected, q.sub.+ (t.sub.2)=0, so EQU P(t.sub.2)=-e/C (6)
Similar effects are observed on the negative electrode.
Thus, the effect of an ionization event in an ionization chamber is completed only after collection of all of the ions, both positive and negative. More importantly, there is no sudden change of potential upon collection of ions, but rather, an increasing amount of charge is smoothly induced on the plates as the positive and negative ions approach the plates. See Wilkinson, Ionization Chambers and Counters .sctn.4.2, 59-62 (Cambridge Press 1950).
The desirability of having an ionization chamber in which the output pulse does not depend on the position of uncollected ions in the chamber was recognized long ago. One way to accomplish this result is to place a grid of parallel wires having a spacing .xi. (axis to axis) with wire radius r a distance c from the electron collection electrode of the ionization chamber and a distance a from the other electrode in the chamber. Consider now an ion par formed a distance b from the plane of the grid. The charge induced on the electron collection electrode is no longer ##EQU3## as in the no-grid case, since some of the lines of force produced by the ions finish on the grid instead of on the collection plate. Thus, the electron collection electrode is shielded from the effect of the positive ions. The electron proceeds to the electron collection plate (assuming it is not collected by the grid) and finally produces a charge on the collection plate when it is collected.
Such a grid shields the collection electrode from the effects of charged particles between the grid and the other electrode. Grid shielding efficiency depends on r (the radius of the grid wires), .xi. (the wire spacing), and c (the distance from the grid to the shielded collection electrode).
Charged particles traveling between the grid and the collection electrode induce on the collection electrode a charge equal to their own (i.e. there is a space between the grid and the collection electrode where no shielding action occurs). Nevertheless, vastly increased rise times of the pulses measured at the collection electrode have been observed in ionization chambers including such grids or similar shielding structures. See, e.g., Wilkinson at 74-77; Rossi et al, Ionization Chambers and Counters: Experimental Techniques, Chapter 3 at 31-71 (McGraw-Hill 1949); Bunemann et al, "Design of Grid Ionization Chambers", A27 Can. Journal of Research 191-206 (1949); O. R. Frisch, Unpublished Report BR-49 (British Atomic Energy Project); and U.S. Pat. No. 4,047,040 to Houston (1977).
Gridded ionization chambers are presently in wide use for many applications involving detection and/or identification of charged particles. For instance, U.S. Pat. No. 4,150,290 to Erskine et al (1979) discloses a gridded ionization chamber adapted for detecting the energy, loss of energy per unit distance and angle of incidence of heavy ions. Butz-Jorgensen et al, "Investigation Of Fission Layers For Precise Fission Cross-Section Measurements With A Gridded Ionization Chamber", 86 Nuclear Science and Engineering 10-21 (1984) teaches using an ionization chamber with a Frisch grid to determine both the energy and the emission angle of charged particles emitted from a source positioned co-planar with the cathode of the chamber. Asselineau et al, "Performance of a Bragg Curve Detector For Heavy Ion Identification", 204 Nuclear Instruments and Methods 109-15 (1982) discloses an ionization chamber which continuously samples the ionization along the track left by an entering ion (the sampling being achieved in a short section of the detector defined by a Frisch grid). The atomic number and energy of high energy heavy ions stopping in the ionization chamber are determined by Bragg curve spectroscopy. See also Hotzl et al, "Experiences With Large-Area Frisch Grid Chambers In Low-Level Alpha Spectrometry", 22 Nuclear Instruments and Methods in Physics Research 290-94 (1981) (the use of parallel-plate gridded ionization chambers for alpha spectrometry). Zurmuhle et al, 203 Nuclear Instruments and Methods 261-67 (1982) discloses a heavy ion charged particle telescope using gas ionization chambers with and without Frisch Grids as .DELTA.-E counters. See also Berceanu et al, "Detection And Identification of Heavy Ions at 180.degree. Using a Proportional .DELTA.E-E Chamber", 35 Stud. And Cercet. Fiz. No. 5, 503-505 (Rumania 1983) (cylindrical ionization chamber with Frisch Grid used as a proportional .DELTA.E-E chamber to measure specific energy loss and residual energy of heavy ions).
Other applications of gas ionization chambers include those described in U.S. Pat. No. 4,378,499 to Spangler et al (1983) (ion mobility detectors), U.S. Pat. No. 4,239,967 to Carr et al (1980) (trace water measurement) and U.S. Pat. No. 4,311,908 to Goulianos et al (1982) (gel electrophoresis). Ionization chambers are useful in almost any application wherein some property of an ionization event is to be determined, observed, or measured.
There have been two proposed methods for adjusting or compensating an x-ray beam depending upon patient thickness. The first is the Plewes method which is normally used in a screen-film system. A detector placed behind the film measures the transmitted intensity and feeds a signal back to the x-ray tube grid, which controls the tube current so as to maintain a more equal intensity of x-radiation across the film. To make use of this method it is necessary to scan a "thick" pencil beam of x-rays (.about.3.times.3 cm) across the film. A disadvantage of the Plewes method is that it is fairly slow, requiring about 1-5 seconds for a chest exposure. Accordingly, it is subject to motion artifacts and has all the deficiencies of a film-screen system.
The second method is the Mistretta (Wisconsin) method, where a preliminary low-dose exposure of the patient is made from which a mask filter with spatially-varying thickness is made. This filter is placed in the beam near the focal spot and a second, diagnostic exposure is made having a more nearly equalized intensity across the film. A disadvantage of this method is the time required to make and position the filter mask and that patient motion could occur during this time. The method also suffers from all the deficiencies of film-screens.